AI billing tools helped add nearly $1 billion in extra hospital charges, insurer says
Blue Cross Blue Shield says AI-assisted hospital coding helped add nearly $1 billion in extra costs, often with no change in patient care.
What happened: Blue Cross Blue Shield Association, the insurance trade group covering roughly one-third of Americans, says AI-assisted medical coding at hospitals contributed to about $942 million in extra costs across its health plans between 2023 and 2025. The bulk of that increase came from secondary diagnoses that moved patients into higher-paying billing categories. BCBSA says roughly $653 million of that total, about 70 percent, involved additional diagnoses that were not matched by any change in the patient's actual treatment. The report lands as 60 percent of hospital systems have adopted AI coding tools, and as hospitals and insurers each build out AI systems to review, approve, and contest the same claims.
Why it matters: AI was supposed to make healthcare paperwork cheaper by catching errors and automating busywork. Instead, early evidence suggests it may be making the system's existing billing incentives easier to exploit. More complex coding can mean bigger reimbursements without any more care being delivered, and those costs tend to flow downstream to patients through higher premiums, deductibles, and taxes. Benefits consultant Marsh projects employer health coverage costs will rise 8.2 percent in 2027, the steepest increase since 2003. Economists warn of an administrative arms race, where hospitals deploy AI to capture more billable diagnoses, insurers deploy AI to deny or downcode claims, and any savings get spent on dueling software instead.
How it works, plainly: Hospitals already earn more for patients with multiple documented conditions, called secondary diagnoses, because sicker patients in theory cost more to treat. AI tools now scan charts and lab results to flag conditions clinicians may not have written down, sometimes from a single abnormal lab value. That can push a patient into a higher-paying billing bracket even when it has no bearing on the care actually given. On the other side, insurers run their own AI to scrutinize claims and flag ones for denial or reduced payment, while hospitals use AI to draft appeals faster. One health economist described this as AI making it easier to capture incentives that already existed in billing, not inventing new ones.
What comes next: The American Hospital Association rejects BCBSA's framing, arguing patients today are older and sicker and that AI simply helps providers document real conditions more accurately. It calls the insurer's analysis incomplete and accuses insurers of running their own AI-driven denial and downcoding practices. Industry voices do agree on one fix: keep a human reviewing AI-flagged diagnoses and denials instead of letting software decide alone, and audit AI-generated coding the way human coders have long been audited. Whether regulators intervene, or hospitals and insurers simply keep out-spending each other on competing AI systems, will decide if the technology lowers costs or just shifts who wins each billing dispute.